In this chapter, we introduced you to different convolutional neural network designs that have proven their effectiveness and, as a result, are widely used. We started by introducing the VGGNet model by VGG at Oxford University. Next, we moved on to GoogLeNet by Google, before finally talking about Microsoft's Residual Net. In addition, we showed you a more advanced and new type of convolution that is featured in a model design called MobileNet. Throughout, we talked about the different properties and design choices that make each of these networks so good, such as skip connections, stacking small filters, or inception modules. Finally, code was given showing you how to write out these networks in TensorFlow.

In the next chapter, we will talk about a new kind of model, called a generative model, which will allow us to generate data.